Robotic Blossom Thinning System for Tree Fruit Crops

Uddhav Bhattarai

doi:10.7273/000005022

Tree fruit production industry around the world heavily relies on semi-skilled seasonal workforce for critical field operations such as training, pruning, blossom and fruitlet thinning, and harvesting. Blossom thinning is an essential crop-load management technique that relies heavily on laborious and labor-intensive manual operation to achieve the desired thinning result. While large-scale thinning approaches such as chemical and mechanical thinning are available, chemical thinning results can be unpredictable, and mechanical thinning may damage a significant part of the tree canopy while also offering no-to-limited selectivity. Therefore, developing an efficient system that can perform precise blossom thinning in the target canopy regions with high accuracy, effectiveness, and robustness is crucial. This study focused on the design, development, and field evaluation of a robotic blossom thinning system that employed a machine vision system, and a mechatronic system consisting of a robotic manipulator and end-effector for targeted, selective blossom thinning in tree fruit crops. Robust machine vision systems were investigated for the identification, segmentation, density estimation, localization, and counting of apple flowers. Furthermore, a miniature, electrically-actuated end-effector was designed, fabricated, and tested for blossom thinning in space-constrained locations in tree canopies. All these components were integrated to develop a robotic thinning system and evaluated in a commercial orchard. Two thinning methods, boundary and center thinning, were investigated to assess the integrated system performance in selectively thinning flower clusters in target canopy regions. Boundary thinning was used to thin flowers along the flower cluster boundary, whereas center thinning was used to thin flowers by actuating the thinning end-effector at the center of the target cluster. The field evaluation results demonstrated that the integrated system could selectively thin blossoms from targeted clusters based on the chosen thinning strategy. The boundary thinning approach achieved a 67.2% thinning with a cycle time of 9.0 seconds, whereas the center thinning approach thinned 59.4% of flowers with a cycle time of 7.2 seconds per cluster. When implemented at a wider scale with additional improvements, the proposed system could address the problems associated with current hand, chemical, and mechanical blossom thinning approaches. Furthermore, the proposed system could aid in the commercial viability and practical adoption of the robotic systems intended for operation in tree fruit crops.

Robotic Blossom Thinning System for Tree Fruit Crops

Files and links (1)

Abstract

Metrics

Details